Repository contains finished project completed during Compilers course at AGH UST.
Statically-typed Matlab-like language compiled using LLVM compiler infrastracture.
See Usage example below to see generated ir for this code.
Code:
M = zeros(3, 3);
print M;
# Assigining individual values works!
M[1, 1] = 69.0;
# Reading values works!
x = M[1, 1];
print "x =", x;
X = [1, 2, 3];
for i = 0:2 {
M[i, i] = X[i];
}
print M;
Output:
[ 0.00, 0.00, 0.00 ]
[ 0.00, 0.00, 0.00 ]
[ 0.00, 0.00, 0.00 ]
x = 69.00
[ 1.00, 0.00, 0.00 ]
[ 0.00, 2.00, 0.00 ]
[ 0.00, 0.00, 3.00 ]
Makefile prints AST, both unoptimized and optimized IR and generates executable file which is automatically executed.
Command:
make all arg=./tests/ir/slice.m
Output:
------------------------ Compiling Runtime -------------------------
clang++ -shared -fpic -lpthread runtime.cpp -o ./build/runtime.so
------------------------ Compiler -------------------------
<Target x86-64 (64-bit X86: EM64T and AMD64)>
------------------------ Compiling ./tests/ir/slice.m ------------------------
M = zeros(3, 3);
print M;
# Assigining individual values works!
M[1, 1] = 69.0;
# Reading values works!
x = M[1, 1];
print "x =", x;
X = [1, 2, 3];
for i = 0:2 {
M[i, i] = X[i];
}
print M;
------------------------ Abstract Syntax Tree ------------------------
Root
└── CodeBlock
├── =
│ ├── FunctionCall
│ │ ├── Int
│ │ │ └── 3
│ │ ├── Int
│ │ │ └── 3
│ │ └── zeros
│ └── ID
│ └── M
├── =
│ ├── Float
│ │ └── 69.0
│ └── VectorSlice
│ ├── ID
│ │ └── M
│ ├── SimpleRange
│ │ └── Int
│ │ └── 1
│ └── SimpleRange
│ └── Int
│ └── 1
├── =
│ ├── ID
│ │ └── x
│ └── VectorSlice
│ ├── ID
│ │ └── M
│ ├── SimpleRange
│ │ └── Int
│ │ └── 1
│ └── SimpleRange
│ └── Int
│ └── 1
├── =
│ ├── ID
│ │ └── X
│ └── Vector
│ ├── Int
│ │ └── 1
│ ├── Int
│ │ └── 2
│ └── Int
│ └── 3
├── For
│ ├── CodeBlock
│ │ └── =
│ │ ├── VectorSlice
│ │ │ ├── ID
│ │ │ │ └── M
│ │ │ ├── SimpleRange
│ │ │ │ └── ID
│ │ │ │ └── i
│ │ │ └── SimpleRange
│ │ │ └── ID
│ │ │ └── i
│ │ └── VectorSlice
│ │ ├── ID
│ │ │ └── X
│ │ └── SimpleRange
│ │ └── ID
│ │ └── i
│ ├── ID
│ │ └── i
│ └── Range
│ ├── Int
│ │ └── 0
│ └── Int
│ └── 2
├── FunctionCall
│ ├── ID
│ │ └── M
│ └── print
├── FunctionCall
│ ├── ID
│ │ └── x
│ ├── String
│ │ └── x =
│ └── print
└── FunctionCall
├── ID
│ └── M
└── print
------------------------ Unoptimized IR ------------------------
; ModuleID = "Main"
target triple = "unknown-unknown-unknown"
target datalayout = ""
declare i8* @"formatInt"(i32 %".1")
declare i8* @"formatDouble"(double %".1")
declare void @"freeString"(i8* %".1")
declare void @"putLn"()
declare void @"putStrLn"(i8* %".1")
declare void @"putVectorLn"({}* %".1")
declare void @"putStr"(i8* %".1")
declare {}* @"zeros"(i32 %".1", i32* %".2")
declare {}* @"ones"(i32 %".1", i32* %".2")
declare {}* @"dotAdd"({}* %".1", {}* %".2")
declare {}* @"dotMinus"({}* %".1", {}* %".2")
declare {}* @"dotMult"({}* %".1", {}* %".2")
declare {}* @"dotDiv"({}* %".1", {}* %".2")
declare void @"assignValue"({}* %".1", i32* %".2", i32 %".3", double %".4")
declare double @"readValue"({}* %".1", i32* %".2", i32 %".3")
declare {}* @"literalNVector"(i32 %".1", i32* %".2", double* %".3")
define i32 @"main"()
{
entry:
%".2" = alloca [2 x i32]
%".3" = getelementptr [2 x i32], [2 x i32]* %".2", i32 0, i32 0
store i32 3, i32* %".3"
%".5" = getelementptr [2 x i32], [2 x i32]* %".2", i32 0, i32 1
store i32 3, i32* %".5"
%".7" = getelementptr [2 x i32], [2 x i32]* %".2", i32 0, i32 0
%".8" = call {}* @"zeros"(i32 2, i32* %".7")
%"M" = alloca {}*
store {}* %".8", {}** %"M"
%".10" = load {}*, {}** %"M"
call void @"putVectorLn"({}* %".10")
call void @"putLn"()
%".13" = alloca [6 x i32]
%".14" = getelementptr [6 x i32], [6 x i32]* %".13", i32 0, i32 0
store i32 1, i32* %".14"
%".16" = getelementptr [6 x i32], [6 x i32]* %".13", i32 0, i32 1
store i32 1, i32* %".16"
%".18" = getelementptr [6 x i32], [6 x i32]* %".13", i32 0, i32 2
store i32 0, i32* %".18"
%".20" = getelementptr [6 x i32], [6 x i32]* %".13", i32 0, i32 3
store i32 1, i32* %".20"
%".22" = getelementptr [6 x i32], [6 x i32]* %".13", i32 0, i32 4
store i32 1, i32* %".22"
%".24" = getelementptr [6 x i32], [6 x i32]* %".13", i32 0, i32 5
store i32 0, i32* %".24"
%".26" = load {}*, {}** %"M"
%".27" = getelementptr [6 x i32], [6 x i32]* %".13", i32 0, i32 0
call void @"assignValue"({}* %".26", i32* %".27", i32 6, double 0x4051400000000000)
%".29" = alloca [6 x i32]
%".30" = getelementptr [6 x i32], [6 x i32]* %".29", i32 0, i32 0
store i32 1, i32* %".30"
%".32" = getelementptr [6 x i32], [6 x i32]* %".29", i32 0, i32 1
store i32 1, i32* %".32"
%".34" = getelementptr [6 x i32], [6 x i32]* %".29", i32 0, i32 2
store i32 0, i32* %".34"
%".36" = getelementptr [6 x i32], [6 x i32]* %".29", i32 0, i32 3
store i32 1, i32* %".36"
%".38" = getelementptr [6 x i32], [6 x i32]* %".29", i32 0, i32 4
store i32 1, i32* %".38"
%".40" = getelementptr [6 x i32], [6 x i32]* %".29", i32 0, i32 5
store i32 0, i32* %".40"
%".42" = load {}*, {}** %"M"
%".43" = getelementptr [6 x i32], [6 x i32]* %".29", i32 0, i32 0
%".44" = call double @"readValue"({}* %".42", i32* %".43", i32 6)
%"x" = alloca double
store double %".44", double* %"x"
call void @"putStr"(i8* getelementptr ([4 x i8], [4 x i8]* @"global_0", i32 0, i32 0))
%".47" = load double, double* %"x"
%".48" = call i8* @"formatDouble"(double %".47")
call void @"putStr"(i8* %".48")
call void @"freeString"(i8* %".48")
call void @"putLn"()
%".52" = sitofp i32 1 to double
%".53" = getelementptr [3 x double], [3 x double]* @"global_1", i32 0, i32 0
store double %".52", double* %".53"
%".55" = sitofp i32 2 to double
%".56" = getelementptr [3 x double], [3 x double]* @"global_1", i32 0, i32 1
store double %".55", double* %".56"
%".58" = sitofp i32 3 to double
%".59" = getelementptr [3 x double], [3 x double]* @"global_1", i32 0, i32 2
store double %".58", double* %".59"
%".61" = call {}* @"literalNVector"(i32 1, i32* getelementptr ([1 x i32], [1 x i32]* @"global_2", i32 0, i32 0), double* getelementptr ([3 x double], [3 x double]* @"global_1", i32 0, i32 0))
%"X" = alloca {}*
store {}* %".61", {}** %"X"
br label %"for-init-block"
for-init-block:
%"i" = alloca i32
store i32 0, i32* %"i"
br label %"for-condition"
for-condition:
%".66" = load i32, i32* %"i"
%".67" = sitofp i32 %".66" to double
%".68" = sitofp i32 2 to double
%"for-cmp" = fcmp ule double %".67", %".68"
br i1 %"for-cmp", label %"for-body-block", label %"for-merged"
for-body-block:
%".70" = alloca [3 x i32]
%".71" = load i32, i32* %"i"
%".72" = getelementptr [3 x i32], [3 x i32]* %".70", i32 0, i32 0
store i32 %".71", i32* %".72"
%".74" = getelementptr [3 x i32], [3 x i32]* %".70", i32 0, i32 1
store i32 %".71", i32* %".74"
%".76" = getelementptr [3 x i32], [3 x i32]* %".70", i32 0, i32 2
store i32 0, i32* %".76"
%".78" = load {}*, {}** %"X"
%".79" = getelementptr [3 x i32], [3 x i32]* %".70", i32 0, i32 0
%".80" = call double @"readValue"({}* %".78", i32* %".79", i32 3)
%".81" = alloca [6 x i32]
%".82" = load i32, i32* %"i"
%".83" = getelementptr [6 x i32], [6 x i32]* %".81", i32 0, i32 0
store i32 %".82", i32* %".83"
%".85" = getelementptr [6 x i32], [6 x i32]* %".81", i32 0, i32 1
store i32 %".82", i32* %".85"
%".87" = getelementptr [6 x i32], [6 x i32]* %".81", i32 0, i32 2
store i32 0, i32* %".87"
%".89" = load i32, i32* %"i"
%".90" = getelementptr [6 x i32], [6 x i32]* %".81", i32 0, i32 3
store i32 %".89", i32* %".90"
%".92" = getelementptr [6 x i32], [6 x i32]* %".81", i32 0, i32 4
store i32 %".89", i32* %".92"
%".94" = getelementptr [6 x i32], [6 x i32]* %".81", i32 0, i32 5
store i32 0, i32* %".94"
%".96" = load {}*, {}** %"M"
%".97" = getelementptr [6 x i32], [6 x i32]* %".81", i32 0, i32 0
call void @"assignValue"({}* %".96", i32* %".97", i32 6, double %".80")
%".99" = load i32, i32* %"i"
%".100" = add i32 %".99", 1
store i32 %".100", i32* %"i"
br label %"for-condition"
for-merged:
%".103" = load {}*, {}** %"M"
call void @"putVectorLn"({}* %".103")
call void @"putLn"()
ret i32 0
}
@"global_0" = constant [4 x i8] [i8 120, i8 32, i8 61, i8 0]
@"global_1" = global [3 x double] [double 0x0, double 0x0, double 0x0]
@"global_2" = constant [1 x i32] [i32 3]
------------------------ Optimized IR ------------------------
; ModuleID = '<string>'
source_filename = "<string>"
target triple = "unknown-unknown-unknown"
@global_0 = constant [4 x i8] c"x =\00"
@global_1 = global [3 x double] zeroinitializer
@global_2 = constant [1 x i32] [i32 3]
declare i8* @formatDouble(double) local_unnamed_addr
declare void @freeString(i8*) local_unnamed_addr
declare void @putLn() local_unnamed_addr
declare void @putVectorLn({}*) local_unnamed_addr
declare void @putStr(i8*) local_unnamed_addr
declare {}* @zeros(i32, i32*) local_unnamed_addr
declare void @assignValue({}*, i32*, i32, double) local_unnamed_addr
declare double @readValue({}*, i32*, i32) local_unnamed_addr
declare {}* @literalNVector(i32, i32*, double*) local_unnamed_addr
define i32 @main() local_unnamed_addr {
entry:
%.2 = alloca [2 x i32], align 4
%.3 = getelementptr inbounds [2 x i32], [2 x i32]* %.2, i64 0, i64 0
store i32 3, i32* %.3, align 4
%.5 = getelementptr inbounds [2 x i32], [2 x i32]* %.2, i64 0, i64 1
store i32 3, i32* %.5, align 4
%.8 = call {}* @zeros(i32 2, i32* nonnull %.3)
call void @putVectorLn({}* %.8)
call void @putLn()
%.13 = alloca [6 x i32], align 4
%.14 = getelementptr inbounds [6 x i32], [6 x i32]* %.13, i64 0, i64 0
store i32 1, i32* %.14, align 4
%.16 = getelementptr inbounds [6 x i32], [6 x i32]* %.13, i64 0, i64 1
store i32 1, i32* %.16, align 4
%.18 = getelementptr inbounds [6 x i32], [6 x i32]* %.13, i64 0, i64 2
store i32 0, i32* %.18, align 4
%.20 = getelementptr inbounds [6 x i32], [6 x i32]* %.13, i64 0, i64 3
store i32 1, i32* %.20, align 4
%.22 = getelementptr inbounds [6 x i32], [6 x i32]* %.13, i64 0, i64 4
store i32 1, i32* %.22, align 4
%.24 = getelementptr inbounds [6 x i32], [6 x i32]* %.13, i64 0, i64 5
store i32 0, i32* %.24, align 4
call void @assignValue({}* %.8, i32* nonnull %.14, i32 6, double 6.900000e+01)
%.29 = alloca [6 x i32], align 4
%.30 = getelementptr inbounds [6 x i32], [6 x i32]* %.29, i64 0, i64 0
store i32 1, i32* %.30, align 4
%.32 = getelementptr inbounds [6 x i32], [6 x i32]* %.29, i64 0, i64 1
store i32 1, i32* %.32, align 4
%.34 = getelementptr inbounds [6 x i32], [6 x i32]* %.29, i64 0, i64 2
store i32 0, i32* %.34, align 4
%.36 = getelementptr inbounds [6 x i32], [6 x i32]* %.29, i64 0, i64 3
store i32 1, i32* %.36, align 4
%.38 = getelementptr inbounds [6 x i32], [6 x i32]* %.29, i64 0, i64 4
store i32 1, i32* %.38, align 4
%.40 = getelementptr inbounds [6 x i32], [6 x i32]* %.29, i64 0, i64 5
store i32 0, i32* %.40, align 4
%.44 = call double @readValue({}* %.8, i32* nonnull %.30, i32 6)
call void @putStr(i8* getelementptr inbounds ([4 x i8], [4 x i8]* @global_0, i64 0, i64 0))
%.48 = call i8* @formatDouble(double %.44)
call void @putStr(i8* %.48)
call void @freeString(i8* %.48)
call void @putLn()
store double 1.000000e+00, double* getelementptr inbounds ([3 x double], [3 x double]* @global_1, i64 0, i64 0), align 16
store double 2.000000e+00, double* getelementptr inbounds ([3 x double], [3 x double]* @global_1, i64 0, i64 1), align 8
store double 3.000000e+00, double* getelementptr inbounds ([3 x double], [3 x double]* @global_1, i64 0, i64 2), align 16
%.61 = call {}* @literalNVector(i32 1, i32* getelementptr inbounds ([1 x i32], [1 x i32]* @global_2, i64 0, i64 0), double* getelementptr inbounds ([3 x double], [3 x double]* @global_1, i64 0, i64 0))
%.70 = alloca [3 x i32], align 4
%.72 = getelementptr inbounds [3 x i32], [3 x i32]* %.70, i64 0, i64 0
store i32 0, i32* %.72, align 4
%.74 = getelementptr inbounds [3 x i32], [3 x i32]* %.70, i64 0, i64 1
store i32 0, i32* %.74, align 4
%.76 = getelementptr inbounds [3 x i32], [3 x i32]* %.70, i64 0, i64 2
store i32 0, i32* %.76, align 4
%.80 = call double @readValue({}* %.61, i32* nonnull %.72, i32 3)
%.81 = alloca [6 x i32], align 4
%.83 = getelementptr inbounds [6 x i32], [6 x i32]* %.81, i64 0, i64 0
%0 = bitcast [6 x i32]* %.81 to i8*
call void @llvm.memset.p0i8.i64(i8* nonnull align 4 dereferenceable(24) %0, i8 0, i64 24, i1 false)
call void @assignValue({}* %.8, i32* nonnull %.83, i32 6, double %.80)
%.70.1 = alloca [3 x i32], align 4
%.72.1 = getelementptr inbounds [3 x i32], [3 x i32]* %.70.1, i64 0, i64 0
store i32 1, i32* %.72.1, align 4
%.74.1 = getelementptr inbounds [3 x i32], [3 x i32]* %.70.1, i64 0, i64 1
store i32 1, i32* %.74.1, align 4
%.76.1 = getelementptr inbounds [3 x i32], [3 x i32]* %.70.1, i64 0, i64 2
store i32 0, i32* %.76.1, align 4
%.80.1 = call double @readValue({}* %.61, i32* nonnull %.72.1, i32 3)
%.81.1 = alloca [6 x i32], align 4
%.83.1 = getelementptr inbounds [6 x i32], [6 x i32]* %.81.1, i64 0, i64 0
store i32 1, i32* %.83.1, align 4
%.85.1 = getelementptr inbounds [6 x i32], [6 x i32]* %.81.1, i64 0, i64 1
store i32 1, i32* %.85.1, align 4
%.87.1 = getelementptr inbounds [6 x i32], [6 x i32]* %.81.1, i64 0, i64 2
store i32 0, i32* %.87.1, align 4
%.90.1 = getelementptr inbounds [6 x i32], [6 x i32]* %.81.1, i64 0, i64 3
store i32 1, i32* %.90.1, align 4
%.92.1 = getelementptr inbounds [6 x i32], [6 x i32]* %.81.1, i64 0, i64 4
store i32 1, i32* %.92.1, align 4
%.94.1 = getelementptr inbounds [6 x i32], [6 x i32]* %.81.1, i64 0, i64 5
store i32 0, i32* %.94.1, align 4
call void @assignValue({}* %.8, i32* nonnull %.83.1, i32 6, double %.80.1)
%.70.2 = alloca [3 x i32], align 4
%.72.2 = getelementptr inbounds [3 x i32], [3 x i32]* %.70.2, i64 0, i64 0
store i32 2, i32* %.72.2, align 4
%.74.2 = getelementptr inbounds [3 x i32], [3 x i32]* %.70.2, i64 0, i64 1
store i32 2, i32* %.74.2, align 4
%.76.2 = getelementptr inbounds [3 x i32], [3 x i32]* %.70.2, i64 0, i64 2
store i32 0, i32* %.76.2, align 4
%.80.2 = call double @readValue({}* %.61, i32* nonnull %.72.2, i32 3)
%.81.2 = alloca [6 x i32], align 4
%.83.2 = getelementptr inbounds [6 x i32], [6 x i32]* %.81.2, i64 0, i64 0
store i32 2, i32* %.83.2, align 4
%.85.2 = getelementptr inbounds [6 x i32], [6 x i32]* %.81.2, i64 0, i64 1
store i32 2, i32* %.85.2, align 4
%.87.2 = getelementptr inbounds [6 x i32], [6 x i32]* %.81.2, i64 0, i64 2
store i32 0, i32* %.87.2, align 4
%.90.2 = getelementptr inbounds [6 x i32], [6 x i32]* %.81.2, i64 0, i64 3
store i32 2, i32* %.90.2, align 4
%.92.2 = getelementptr inbounds [6 x i32], [6 x i32]* %.81.2, i64 0, i64 4
store i32 2, i32* %.92.2, align 4
%.94.2 = getelementptr inbounds [6 x i32], [6 x i32]* %.81.2, i64 0, i64 5
store i32 0, i32* %.94.2, align 4
call void @assignValue({}* %.8, i32* nonnull %.83.2, i32 6, double %.80.2)
call void @putVectorLn({}* %.8)
call void @putLn()
ret i32 0
}
; Function Attrs: argmemonly nounwind willreturn
declare void @llvm.memset.p0i8.i64(i8* nocapture writeonly, i8, i64, i1 immarg) #0
attributes #0 = { argmemonly nounwind willreturn }
------------------------ Linker -------------------------
clang++ ./build/output.o ./build/runtime.so -o ./build/executable.exe
------------------------ Go! -----------------------------
[ 0.00, 0.00, 0.00 ]
[ 0.00, 0.00, 0.00 ]
[ 0.00, 0.00, 0.00 ]
x = 69.00
[ 1.00, 0.00, 0.00 ]
[ 0.00, 2.00, 0.00 ]
[ 0.00, 0.00, 3.00 ]
- Lexer and Parser (LALR) come from PLY.
- clang - for compiling shared runtime and linking it with object file produced by code generator.
- llvmlite - provides python bindings to llvm.
- llvm 10.0.0 (actually I only use clang)
- ply 3.11
- llvmlite 0.35.0